GPU-Accelerated Monte-Carlo Scatter Correction for PET Sinograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating gamma-ray scattering in positron emission tomography (PET) are inefficient due to high operation times, making accurate 3D image reconstruction challenging, especially with low radiation levels and high noise levels, which limits the clinical application of Monte-Carlo simulations.
Innovation Solution
Employing a graphics processing unit (GPU) for parallel processing of Monte-Carlo simulations to estimate gamma-ray scattering, utilizing a hybrid CPU-GPU structure for generating a gamma-ray scatter sinogram by simulating photon migration and correcting for scattering, thereby reducing operation time and improving image reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Monte-Carlo simulation is used for gamma-ray scattering estimation, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent divides the Monte-Carlo simulation process into multiple independent calculation units that can be executed in parallel. Each processing element handles a portion of the photon migration simulations, allowing the overall computation to be segmented across multiple GPU cores simultaneously, thereby reducing total operation time while maintaining accuracy
Solution Approach 2:
The patent transitions from sequential processing (single dimension of time) to parallel processing by utilizing the spatial architecture of GPU cores. Multiple calculation threads are launched simultaneously across hundreds of GPU cores, adding a parallel dimension to the computation that dramatically reduces operation time from hours to minutes
2Manufacturing precision
If radiation level is increased to improve image quality, then manufacturing precision is improved, but object-affected harmful factors increase
Solution Approach 1:
The patent implements a feedback mechanism where the GPU-accelerated Monte-Carlo simulation predicts gamma-ray scattering patterns, which are then used to correct the reconstructed image. This feedback loop allows the system to compensate for scattering effects mathematically, enabling accurate image reconstruction from low-radiation data without requiring increased radiation exposure
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach significantly reduces operation time for gamma-ray scattering estimation and enables accurate 3D image reconstruction, making Monte-Carlo simulations feasible for clinical use by leveraging the parallel processing capabilities of GPUs.
Implementation Method 1
Monte-Carlo simulation generates a gamma-ray photon one by one and simulates of gamma-ray photon migrations to be similar to the real
Implementation Method 2
scattering of gamma rays may greatly affect an image reconstruction result
Data Source
AI summary
A method and apparatus for reconstructing an image by generating a scatter-corrected sinogram are provided. In the reconstructed image in which initial scattering is not corrected, an emission position of a photon is determined. A graphics processing unit (GPU) calculates migration of the photon using a Monte-Carlo simulation. A 3-dimensional (3D) line of response (LoR) may be calculated through a position in which the photon reaches a detector. The calculated LoR may be converted into a 3D sinogram form and stored in a prompt sinogram and a scatter-corrected sinogram. A final scatter-corrected sinogram may be generated by adjusting a scale of the scatter-corrected sinogram.